memory-maintenance

What This Skill Does

The memory-maintenance skill serves as an automated librarian for your OpenClaw agent, ensuring that your agent's historical data remains organized, searchable, and concise. As agents accumulate daily session notes, their context windows become bloated with repetitive or stale information. This skill systematically reviews daily memory files (memory/YYYY-MM-DD.md), extracts key insights, suggests updates to a centralized MEMORY.md file, and manages file retention to prevent directory clutter. By automating the transition from raw daily notes to a structured knowledge base, it ensures your agent remains performant, intelligent, and context-aware over long-term operations without manual intervention.

Installation

To integrate this skill, use the ClawHub package manager. Run the following command in your terminal: clawhub install openclaw/skills/skills/maxlauriehutchinson/memory-maintenance. Once installed, verify the configuration in config/settings.json. You may adjust the retention period (default 30 days) and the daily review schedule (default 23:00) to match your workflow. You can trigger a manual run at any time using openclaw skill memory-maintenance run.

Use Cases

This skill is essential for agents used in long-running projects, such as continuous software development or personal journaling. It is ideal for users who want to retain a 'long-term memory' of decisions without manually sifting through months of logs. It also works perfectly for agents managing research tasks, where distinct daily progress must be aggregated into a final project report.

Example Prompts

1. "Perform a manual memory review for the last 7 days and summarize any recurring tasks into my MEMORY.md file."
2. "Show me the pending suggestions from today's memory-maintenance report so I can approve the changes."
3. "Apply all safe memory cleanup tasks to archive files older than 30 days."

Tips & Limitations

For best results, use gemini-2.5-flash as it provides the perfect balance of speed and reasoning accuracy for summarizing text. Always perform a --dry-run before applying changes to ensure the AI's logic aligns with your preference for what constitutes a 'keeper' note. Note that this skill requires read and write access to your local memory directory; ensure you have backups of your notes if you are testing custom retention policies for the first time.